[Feature] Introduces V2 programs.

[d0cd]

This PR introduces the syntax for V2 programs.
At a high level, V2 programs:

• start with a program$2 keyword
• allow struct, record, mapping, function, closure, and import.
• introduce an optional constructor via the _init keyword. Constructors must be specified after the above constructs.
• introduce user-defined $metadata to the program. Metadata must be specified after the constructor.

Here is an example of a V2 program.

import eth.aleo;
import usdc.aleo;

program$2 swap.aleo;

function swap:
    input r0 as eth.aleo/eth.record;
    input r1 as usdc.aleo/usdc.record;
    call eth.aleo/transfer r0 r1.owner r0.amount into r2 r3;
    call usdc.aleo/transfer r1 r0.owner r1.amount into r4 r5;
    output r2 as eth.aleo/eth.record;
    output r4 as usdc.aleo/usdc.record;

_init:
    assert.eq true true;
    
$metadata edition: 0u8;
$metadata owner: aleo1rhgdu77hgyqd3xjj8ucu3jj9r2krwz6mnzyd80gncr5fxcwlh5rsvzp9px;
$metadata upgradable: false;

The implementation for V2 programs will be included in a following PR.

Testing

• Regressions checking that V1 programs still (de)serialize correctly can be found here under test_program_serde

[vicsn]

From my humble POV, I'm not a fan of adding a new symbol $ to the syntax - even if we don't typically expose aleo instructions to end users. Also because it doesn't work the same way like it does in the other syntax everyone knows (shell scripts).

• For program version: what about a new keyword like version which we declare after import and before program? Related: I think metadata.version in the PR description should be changed to metadata.edition.
• For metadata we could simply drop the $.

[d0cd]

I don't have strong opinions here. I'll leave it to the reviewers to arbitrate on syntax.
FWIW, I feel like the version keyword doesn't create enough of a conceptual separation but I can be convinced if I stare at some programs long enough.